def testDeleteBeforeIndexMerge(self):
reader = IndexReader.open(self.dir, False)
self.assertEqual(2, reader.maxDoc())
self.assertEqual(2, reader.numDocs())
reader.deleteDocument(1)
deletedDocs = MultiFields.getDeletedDocs(reader)
self.assert_(deletedDocs is not None and deletedDocs.get(1))
self.assert_(reader.hasDeletions())
self.assertEqual(2, reader.maxDoc())
self.assertEqual(1, reader.numDocs())
reader.close()
reader = IndexReader.open(self.dir, True)
self.assertEqual(2, reader.maxDoc())
self.assertEqual(1, reader.numDocs())
reader.close()
def testDeleteAfterIndexMerge(self):
reader = IndexReader.open(self.dir, False)
self.assertEqual(2, reader.maxDoc())
self.assertEqual(2, reader.numDocs())
reader.deleteDocument(1)
reader.close()
writer = IndexWriter(self.dir, self.getAnalyzer(), False,
IndexWriter.MaxFieldLength.UNLIMITED)
writer.commit()
writer.close()
reader = IndexReader.open(self.dir, True)
deletedDocs = MultiFields.getDeletedDocs(reader)
self.assert_(deletedDocs is None or not deletedDocs.get(1))
self.assert_(not reader.hasDeletions())
self.assertEqual(1, reader.maxDoc())
self.assertEqual(1, reader.numDocs())
reader.close()
def main(cls, argv):
indexDir = System.getProperty("index.dir")
directory = SimpleFSDirectory(File(indexDir))
reader = IndexReader.open(directory, True)
blt = BooksLikeThis(reader)
deletedDocs = MultiFields.getDeletedDocs(reader)
for id in xrange(reader.maxDoc()):
if deletedDocs is not None and deletedDocs.get(id):
continue
doc = reader.document(id)
print ''
print doc.get("title").encode('utf-8')
docs = blt.docsLike(id, doc, 10)
if not docs:
print " None like this"
else:
for doc in docs:
print " ->", doc.get("title").encode('utf-8')